forked from fengyang95/pyCFTrackers
-
Notifications
You must be signed in to change notification settings - Fork 0
/
fft_tools.py
27 lines (24 loc) · 1.06 KB
/
fft_tools.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
import numpy as np
def fft2(x):
return np.fft.fft(np.fft.fft(x, axis=1), axis=0).astype(np.complex64)
def ifft2(x):
return np.fft.ifft(np.fft.ifft(x, axis=1), axis=0).astype(np.complex64)
def cifft2(xf):
x = np.real(ifft2(np.fft.ifftshift(np.fft.ifftshift(xf, 0),1))).astype(np.float32)
return x
def cfft2(x):
in_shape = x.shape
if in_shape[0] % 2 == 1 and in_shape[1] % 2 == 1:
xf = np.fft.fftshift(np.fft.fftshift(fft2(x), 0), 1).astype(np.complex64)
else:
out_shape = list(in_shape)
out_shape[0] = in_shape[0] + (in_shape[0] + 1) % 2
out_shape[1] = in_shape[1] + (in_shape[1] + 1) % 2
out_shape = tuple(out_shape)
xf = np.zeros(out_shape, dtype=np.complex64)
xf[:in_shape[0], :in_shape[1]] = np.fft.fftshift(np.fft.fftshift(fft2(x), 0), 1).astype(np.complex64)
if out_shape[0] != in_shape[0]:
xf[-1,:] = np.conj(xf[0,::-1])
if out_shape[1] != in_shape[1]:
xf[:,-1] = np.conj(xf[::-1,0])
return xf[:in_shape[0],:in_shape[1]]